Analysis: the financial impact of the 2019–2020 CMS-HCC risk adjustment model changes

Cotiviti

07/09/2019

Each year, the Centers for Medicare & Medicaid Services (CMS) announces updates to the Medicare risk adjustment program, often including changes to how risk scores are calculated for member populations of Medicare Advantage (MA) plans. These scores are crucial to a plan’s bottom line, as they are meant to reflect the healthcare expenditures of its Medicare beneficiaries and directly impact the risk adjustment revenue the plan receives from CMS. The changes effective for the 2019 and 2020 payment years (PY) encompassed several mandates from the passage of the 21st Century Cures Act, including the addition of several new risk-generating hierarchical condition categories (HCCs) as well as updates to risk score coefficients.

 

To assess the potential impact of these newly included HCCs and updated coefficients, Cotiviti’s data scientists applied the 2020 (V24), 2019 (V23), and 2017 (V22) CMS-HCC models to 2017 claim and encounter data for three MA plans of different sizes. Here, we summarize what we learned and offer our own recommendations to assist MA plans in mitigating the impact of these changes.

 

Major changes to the PY 2019 and PY 2020 models

CMS added four new HCCs to the 2019 risk adjustment model, focusing on inclusions for chronic conditions related to mental health and substance abuse as well as the severity of chronic kidney disease (Table 1). It also expanded the code set for HCC 55 for drug and alcohol dependence to include drug abuse and its related complications. Mental health disorders and substance abuse continue to be on the rise in the United States, with the opioid epidemic being a perfect example. CMS determined that these additional diagnoses expanding substance use disorders, mental health-related disorders, and the severity of chronic kidney disease are clinically meaningful and predict medical expenditures, and that the conditions can be diagnosed definitively.

 

CMS announced that the 2020 model will use the Alternative Payment Condition Count (APCC) model for blended risk score calculations, rather than its originally proposed Payment Condition Count (PCC) model. As required by the 21st Century Cures Act, this model focuses on the number of conditions that an individual beneficiary may have, with the coefficients starting with members who have at least four conditions, and the model adjusting as the number of conditions increases. It also includes additional HCCs for dementia and pressure ulcers.


 

Category

PY 2019 model changes

PY 2020 model changes

Risk-adjustable conditions

Drug Abuse and Drug/Alcohol Dependence

  • Expanded code set for Drug/Alcohol Dependence (HCC 55)
  • New HCC for Uncomplicated Drug Abuse (HCC 56)

    Mental Health

  • Two new HCCs:
    • Reactive and Unspecific Psychosis (HCC 58)
    • Personality Disorders (HCC 60)
  • Relabeled Major Depressive, Bipolar, and Paranoid Disorders to HCC 59

    Chronic Kidney Disease

  • New HCC for CKD Stage 3 (HCC 138)

Dementia

  • Two new HCCs:
    • Dementia with Complications (HCC 51)
    • Dementia without Complication (HCC 52)

      Pressure Ulcers

  • New HCC for Pressure Ulcer of Skin with Partial Thickness Skin Loss (HCC 159)

Normalization factor

Increased to 1.041 (RAPS) and 1.038 (EDS)

Increased to 1.075 (RAPS) and 1.069 (EDS)

Risk score blend

Shifted to 75% RAPS/25% EDS

Shifted to 50% RAPS/50% EDS


Table 1. Summary of PY 2019 and PY 2020 model changes.

Analysis methodology

To conduct this study, Cotiviti analyzed blinded 2017 claim and submissions data for three existing MA plans. Below are the characteristics of the selected plans:

 

  • Plan 1: a large national plan with approximately 132,000 MA lives
  • Plan 2: a large regional plan in the Northwest with approximately 57,000 MA lives
  • Plan 3: a midsize plan in the Southeast with approximately 25,000 MA lives

Next, our data scientists calculated the PY 2018, 2019, and 2020 risk scores using the appropriate RAPS/EDS blend, normalization, and coding intensity factors defined by CMS for each year. Finally, we estimated the revenue impact using CMS’s annual average cost of approximately $9,368 for a member with a risk score of one, as stated in the final 2019 Call Letter.

Findings

Changes in clinical condition prevalence

Table 2 compares the prevalence of members with the newly included and revised HCCs in the 2019 and 2020 models to the expected prevalence of these conditions in the general population, based on several governmental and academic sources.

While the prevalence of drug and alcohol dependence; major depressive, bipolar, and paranoid disorders; and reactive and unspecific psychosis are all in line with industry standards, this is not the case across all the new revenue-generating HCCs. The prevalence of drug abuse, personality disorders, CKD stage 3, and dementia are substantially lower than expected, indicating that these conditions are potentially being underreported or underdocumented to health plans.

 

Condition

Plan 1

Plan 2

Plan 3

Industry Standard

Members with Drug/Alcohol Dependence

7.22%

4.26%

3.20%

1 to 16%

Members with Drug Abuse, Uncomplicated, Except Cannabis

0.25%

0.14%

0.07%

6%

Members with Reactive and Unspecified Psychosis

0.26%

0.20%

0.21%

0.3 to 0.4%

Members with Major Depressive, Bipolar, and Paranoid Disorders

18.83%

16.30%

8.43%

8% for 60+ years

Members with Personality Disorders

0.05%

0.05%

0.03%

1.4%

Members with CKD Stage 3

9.92%

10.24%

12.70%

24% for 60+ years

Members with Dementia (Complicated and Uncomplicated)

6.74%

8.25%

7.26%

13.9% for 71+ years

 

Table 2. HCC prevalence versus industry standard clinical condition prevalence.

 

We also analyzed the total percentage of members who would fall under the APCC model for PY 2020 across the three plans due to their total number of conditions. As noted in Table 3, the model applies to a significant portion of beneficiaries for all three health plans—more than a quarter for the national plan.

 

Payment HCC counts

Plan 1

Plan 2

Plan 3

4

6.9%

7.5%

8.0%

5

6.6%

4.9%

6.0%

6

4.7%

3.2%

3.9%

7

3.1%

2.1%

2.4%

8

1.9%

1.3%

1.5%

9

1.2%

0.7%

0.9%

10+

1.9%

1.1%

1.4%

Total with 4 or more

26.2%

20.8%

24.0%

 

Table 3. Total percentage of members with four or more HCCs under the 2020 risk adjustment model.

  

Risk score and financial impact

With the introduction of new and expanded risk-generating HCCs, CMS also updated the normalization coefficients across all Medicare segments. As a result, risk scores decreased by approximately 1 to 2 percent across all three plans for PY 2019 and approximately 4 to 5 percent for PY 2020 (Table 4), which also puts a substantial amount of risk adjustment revenue in jeopardy.

 

 

Plan 1

Plan 2

Plan 3

Payment Year

2018

2019

2020

2018

2019

2020

2018

2019

2020

RAPS

1.101

1.076

1.042

0.954

0.932

0.903

1.029

1.006

0.974

EDS

1.085

1.122

1.071

0.948

0.980

0.939

1.014

1.023

0.938

Blended

1.099

1.087

1.053

0.953

0.944

0.917

1.027

1.010

0.976

2019/2020 Blended – 2018 Blended (%)

N/A

-1.03%

-4.37%

N/A

-0.94%

-3.77%

N/A

-1.64%

-4.93%

Revenue Impact ($)

N/A

($12.8M)

($54.2M)

N/A

($5.0M)

($20.2M)

N/A

($3.9M)

($11.6M)

 

Table 4. Risk score variances and financial impact.

Note that across all three plans, the disparity between RAPS and EDS risk scores has become less pronounced over time—evidence that many health plans are focused on ensuring congruency between the two scores. With CMS now transitioning to a 50/50 blend between RAPS and EDS for PY 2020, any disparity will have an even greater impact in the years ahead for MA plans that are unable to address this challenge.

 

Recommendations

As they adapt to these changes, we recommend that MA plans take a three-pronged approach to support their risk adjustment programs.

 

Provider education 

With the addition of these new revenue generating HCCs for PY 2019 and PY 2020, health plans should help their providers improve documentation for drug abuse, personality disorders, psychosis, dementia, and pressure ulcers. This can include outreach centered on best charting practices, such as training videos and continuing education seminars. In addition, if your plan (or accountable care organization) is integrated with a provider’s EMR system and the capability is available, then consider configuring the EMR with alerts or prompts to help providers accurately and completely document these conditions.

 

RAPS/EDS reconciliation tools

Disparities between MA plans’ RAPS and EDS risk scores are improving, but continue to invest in additional analytical and reporting tools for operational teams to better reconcile the two encounter domains. These teams should focus on identifying and isolating undocumented or unsubstantiated HCCs, as well as isolating any discrepancies down to the encounter level to understand what is driving the disparity between the scores. As CMS continues to raise the portion of MA plans’ overall risk scores determined by EDS data, this reconciliation will be increasingly critical.

 

HCC documentation 

Make sure that your suspecting and coding activities incorporate the new and expanded risk generating HCCs. For example, all medical record reviews in Q2 2019 and beyond should capture diagnoses for all paid HCCs, including the new additions. Focus in-home prospective assessments on documenting evidence of these new HCCs as well.

 

About the author

 As vice president of product management, Lesley Brown drives the product strategy for Cotiviti’s end-to-end Risk Adjustment solutions. She has more than 30 years’ experience in product and project management in the healthcare and pharmaceutical industries. Before joining Cotiviti, she served as senior vice president of product management at Halfpenny Technologies, a provider of clinical data exchange solutions for healthcare providers. She was also senior vice president of care management products for TriZetto, where she was accountable for the quality, delivery, and support of the software solutions in the its care management portfolio, and she held multiple positions at Alere Health. Lesley is the co-author of 10 U.S. patents and has been published in 16 peer-reviewed scientific and clinical journals.